NFS mounted rootfs possible via PCMCIA NIC ?

NFS mounted rootfs possible via PCMCIA NIC ?

[Andreas Schaufler]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hello list,

I am trying to configure a notebook with a PCMCIA NIC to boot over network. 
(kernel 2.4.20)
In order to accomplish this I passed over the neccessary configuration 
paramters through the boot loader (ip, root, nfsroot)

The problem is: When the kernel is booting it is trying to configure the 
Network interface, before it has been activated.

Thus my question is: Is it possible somehow to change the order by some 
configuration parameter ?!?

Thanks for answers in advance

- -Andreas


-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD4DBQE997P0hFLSFNIrGmsRAjMpAJitzpgKyTJDg6ubFy17u9GA+EZEAJ9kgBKv
k7Euyw1aT6/6ORWhzPjZwg==
=AmIZ
-----END PGP SIGNATURE-----

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Andrew Morton]

Andreas Schaufler wrote:
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello list,
> 
> I am trying to configure a notebook with a PCMCIA NIC to boot over network.
> (kernel 2.4.20)

Nope.  The kernel does the NFS thing before bringing up cardbus.

This patch worked, back in the 2.4.17 days.  It also fixes some
cardbus bugs.  I don't immediately recall what they were.




 drivers/pcmcia/Makefile |    3 +
 drivers/pcmcia/ds.c     |   13 ++-----
 drivers/pcmcia/yenta.c  |   85 ++++++++++++++++++++++--------------------------
 init/main.c             |    3 -
 4 files changed, 45 insertions(+), 59 deletions(-)

--- 24/drivers/pcmcia/ds.c~pcmcia-ip-autoconf	Wed Dec 11 14:30:56 2002
+++ 24-akpm/drivers/pcmcia/ds.c	Wed Dec 11 14:30:56 2002
@@ -968,14 +968,7 @@ int __init init_pcmcia_ds(void)
     return 0;
 }
 
-#ifdef MODULE
-
-int __init init_module(void)
-{
-    return init_pcmcia_ds();
-}
-
-void __exit cleanup_module(void)
+void __exit cleanup_pcmcia_ds(void)
 {
     int i;
 #ifdef CONFIG_PROC_FS
@@ -990,4 +983,6 @@ void __exit cleanup_module(void)
     kfree(socket_table);
 }
 
-#endif
+module_init(init_pcmcia_ds);
+module_exit(cleanup_pcmcia_ds);
+
--- 24/drivers/pcmcia/Makefile~pcmcia-ip-autoconf	Wed Dec 11 14:30:56 2002
+++ 24-akpm/drivers/pcmcia/Makefile	Wed Dec 11 14:30:56 2002
@@ -22,7 +22,7 @@ ifeq ($(CONFIG_CARDBUS),y)
 endif
 
 ifeq ($(CONFIG_PCMCIA),y)
-  obj-y   := cistpl.o rsrc_mgr.o bulkmem.o ds.o cs.o
+  obj-y   := cistpl.o rsrc_mgr.o bulkmem.o cs.o
   ifeq ($(CONFIG_CARDBUS),y)
     obj-y += cardbus.o yenta.o pci_socket.o
   endif
@@ -38,6 +38,7 @@ ifeq ($(CONFIG_PCMCIA),y)
   ifeq ($(CONFIG_HD64465_PCMCIA),y)
     obj-y += hd64465_ss.o
   endif
+  obj-y   += ds.o
 else
   ifeq ($(CONFIG_PCMCIA),m)
     obj-m   := pcmcia_core.o ds.o
--- 24/drivers/pcmcia/yenta.c~pcmcia-ip-autoconf	Wed Dec 11 14:30:56 2002
+++ 24-akpm/drivers/pcmcia/yenta.c	Wed Dec 11 14:30:56 2002
@@ -574,38 +574,6 @@ static void yenta_get_socket_capabilitie
 	printk("Yenta IRQ list %04x, PCI irq%d\n", socket->cap.irq_mask, socket->cb_irq);
 }
 
-extern void cardbus_register(pci_socket_t *socket);
-
-/*
- * 'Bottom half' for the yenta_open routine. Allocate the interrupt line
- *  and register the socket with the upper layers.
- */
-static void yenta_open_bh(void * data)
-{
-	pci_socket_t * socket = (pci_socket_t *) data;
-
-	/* It's OK to overwrite this now */
-	socket->tq_task.routine = yenta_bh;
-
-	if (!socket->cb_irq || request_irq(socket->cb_irq, yenta_interrupt, SA_SHIRQ, socket->dev->name, socket)) {
-		/* No IRQ or request_irq failed. Poll */
-		socket->cb_irq = 0; /* But zero is a valid IRQ number. */
-		socket->poll_timer.function = yenta_interrupt_wrapper;
-		socket->poll_timer.data = (unsigned long)socket;
-		socket->poll_timer.expires = jiffies + HZ;
-		add_timer(&socket->poll_timer);
-	}
-
-	/* Figure out what the dang thing can do for the PCMCIA layer... */
-	yenta_get_socket_capabilities(socket, isa_interrupts);
-	printk("Socket status: %08x\n", cb_readl(socket, CB_SOCKET_STATE));
-
-	/* Register it with the pcmcia layer.. */
-	cardbus_register(socket);
-
-	MOD_DEC_USE_COUNT;
-}
-
 static void yenta_clear_maps(pci_socket_t *socket)
 {
 	int i;
@@ -823,6 +791,8 @@ static struct cardbus_override_struct {
 
 #define NR_OVERRIDES (sizeof(cardbus_override)/sizeof(struct cardbus_override_struct))
 
+extern void cardbus_register(pci_socket_t *socket);
+
 /*
  * Initialize a cardbus controller. Make sure we have a usable
  * interrupt, and that we can map the cardbus area. Fill in the
@@ -832,15 +802,19 @@ static int yenta_open(pci_socket_t *sock
 {
 	int i;
 	struct pci_dev *dev = socket->dev;
+	int retval = -1;
+	int polling = 0;
+
+	MOD_INC_USE_COUNT;
 
 	/*
 	 * Do some basic sanity checking..
 	 */
 	if (pci_enable_device(dev))
-		return -1;
+		goto fail;
 	if (!pci_resource_start(dev, 0)) {
 		printk("No cardbus resource!\n");
-		return -1;
+		goto fail;
 	}
 
 	/*
@@ -849,7 +823,7 @@ static int yenta_open(pci_socket_t *sock
 	 */
 	socket->base = ioremap(pci_resource_start(dev, 0), 0x1000);
 	if (!socket->base)
-		return -1;
+		goto fail;
 
 	yenta_config_init(socket);
 
@@ -867,24 +841,43 @@ static int yenta_open(pci_socket_t *sock
 		if (dev->vendor == d->vendor && dev->device == d->device) {
 			socket->op = d->op;
 			if (d->op->open) {
-				int retval = d->op->open(socket);
-				if (retval < 0)
-					return retval;
+				int ret = d->op->open(socket);
+				if (ret < 0) {
+					iounmap(socket->base);
+					retval = ret;
+					goto fail;
+				}
 			}
 		}
 	}
 
-	/* Get the PCMCIA kernel thread to complete the
-	   initialisation later. We can't do this here,
-	   because, er, because Linus says so :)
-	*/
-	socket->tq_task.routine = yenta_open_bh;
+	socket->tq_task.routine = yenta_bh;
 	socket->tq_task.data = socket;
 
-	MOD_INC_USE_COUNT;
-	schedule_task(&socket->tq_task);
+	if (!socket->cb_irq || request_irq(socket->cb_irq, yenta_interrupt,
+				SA_SHIRQ, socket->dev->name, socket)) {
+		/* No IRQ or request_irq failed. Poll */
+		socket->cb_irq = 0; /* But zero is a valid IRQ number. */
+		socket->poll_timer.function = yenta_interrupt_wrapper;
+		socket->poll_timer.data = (unsigned long)socket;
+		socket->poll_timer.expires = jiffies + HZ;
+		polling = 1;
+	}
 
-	return 0;
+	/* Figure out what the dang thing can do for the PCMCIA layer... */
+	yenta_get_socket_capabilities(socket, isa_interrupts);
+	printk("Socket status: %08x\n", cb_readl(socket, CB_SOCKET_STATE));
+
+	/* Register it with the pcmcia layer.. */
+	cardbus_register(socket);
+
+	if (polling)
+		add_timer(&socket->poll_timer);
+
+	retval = 0;
+fail:
+	MOD_DEC_USE_COUNT;
+	return retval;
 }
 
 /*
--- 24/init/main.c~pcmcia-ip-autoconf	Wed Dec 11 14:30:56 2002
+++ 24-akpm/init/main.c	Wed Dec 11 14:30:56 2002
@@ -533,9 +533,6 @@ static void __init do_basic_setup(void)
 	irda_proto_init();
 	irda_device_init(); /* Must be done after protocol initialization */
 #endif
-#ifdef CONFIG_PCMCIA
-	init_pcmcia_ds();		/* Do this last */
-#endif
 }
 
 extern void prepare_namespace(void);

_

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Andreas Schaufler]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1


> > Hello list,
> >
> > I am trying to configure a notebook with a PCMCIA NIC to boot over
> > network. (kernel 2.4.20)
>
> Nope.  The kernel does the NFS thing before bringing up cardbus.
>
> This patch worked, back in the 2.4.17 days.  It also fixes some
> cardbus bugs.  I don't immediately recall what they were.


I got the 2.4.17 sources and applied the patch. yenta.c and main.c could not 
be patched automatically, so I tried to apply it by hand line by line.
Unfortunately when I boot a kernel compiled witch this modified sources I get 
an "Unable to handle kernel pagin request at virtual address 0000413d"

Maybe this patch is to be used on some Pre Version of 2.4.17 ?!?!

regards
- -Andreas
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.0.7 (GNU/Linux)

iD8DBQE9+MeFhFLSFNIrGmsRAg0HAJ0RBZ/WKDxvv8YXXTLbT7REnqoVowCcC8tL
sGgZre0DIZOAdFVW6kn56rg=
=23D5
-----END PGP SIGNATURE-----

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Andrew Morton]

Andreas Schaufler wrote:
> 
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> > > Hello list,
> > >
> > > I am trying to configure a notebook with a PCMCIA NIC to boot over
> > > network. (kernel 2.4.20)
> >
> > Nope.  The kernel does the NFS thing before bringing up cardbus.
> >
> > This patch worked, back in the 2.4.17 days.  It also fixes some
> > cardbus bugs.  I don't immediately recall what they were.
> 
> I got the 2.4.17 sources and applied the patch. yenta.c and main.c could not
> be patched automatically, so I tried to apply it by hand line by line.
> Unfortunately when I boot a kernel compiled witch this modified sources I get
> an "Unable to handle kernel pagin request at virtual address 0000413d"
> 
> Maybe this patch is to be used on some Pre Version of 2.4.17 ?!?!
> 

Sorry, that patch was against 2.4.20.  It was last tested in 2.4.17.

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Andreas Schaufler]

Am Donnerstag, 12. Dezember 2002 18:35 schrieb Andrew Morton:
...
> >
> > Maybe this patch is to be used on some Pre Version of 2.4.17 ?!?!
>
> Sorry, that patch was against 2.4.20.  It was last tested in 2.4.17.

Ok, thank you very much. It did not patch automatically in 2.4.20 though. But 
after I applied it manually everything works fine.

Now I can use my old Notebook, which has a broken IDE Controller, again :-)

best regards from a happy Linux User
-Andreas

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Alan Cox]

On Wed, 2002-12-11 at 21:53, Andreas Schaufler wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Hello list,
> 
> I am trying to configure a notebook with a PCMCIA NIC to boot over network. 
> (kernel 2.4.20)
> In order to accomplish this I passed over the neccessary configuration 
> paramters through the boot loader (ip, root, nfsroot)
> 
> The problem is: When the kernel is booting it is trying to configure the 
> Network interface, before it has been activated.

PCMCIA relies in part on user space. You can do this, it involves
building a large initrd with a dhcp client on it that sets up pcmcia,
then nfs mounts stuff, then pivot_root()'s into it. Its not exactly
trivial

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[Andreas Schaufler]

...
> PCMCIA relies in part on user space. You can do this, it involves
> building a large initrd with a dhcp client on it that sets up pcmcia,
> then nfs mounts stuff, then pivot_root()'s into it. Its not exactly
> trivial

Thanks for your reply. I get the basic idea from what you say. But what do you 
mean by pivot_root()'ing into it ?!?

I'll try Andrew Morton's suggestion first, because it sounds easier. If I 
can't get it running I'll try your suggestion.

regards
-Andreas

Re: NFS mounted rootfs possible via PCMCIA NIC ?

[H. Peter Anvin]

Followup to:  <200212121829.46465.andreas.schaufler@gmx.de>
By author:    Andreas Schaufler <andreas.schaufler@gmx.de>
In newsgroup: linux.dev.kernel
>
> ...
> > PCMCIA relies in part on user space. You can do this, it involves
> > building a large initrd with a dhcp client on it that sets up pcmcia,
> > then nfs mounts stuff, then pivot_root()'s into it. Its not exactly
> > trivial
> 
> Thanks for your reply. I get the basic idea from what you say. But what do you 
> mean by pivot_root()'ing into it ?!?
> 

pivot_root() is a system call which flips the root directory around.
See Documentation/initrd.txt.

	-hpa
-- 
<hpa@transmeta.com> at work, <hpa@zytor.com> in private!
"Unix gives you enough rope to shoot yourself in the foot."
http://www.zytor.com/~hpa/puzzle.txt	<amsp@zytor.com>